- Joined
- Dec 27, 2006
- Messages
- 54
JD, did you just try to IM me? I got a message about an IM, clicked on the choice to open in a new window, and absolutely nothing happened.
Then you need to decide if the image needs pretty much continuous sampling or if it can tolerate "holes". By holes, I mean whether each exposure has to be right on the tail of another, like two second exposures at two second intervals, or whether there can be gaps, like 1/60 sec exposures at 2 second intervals, or 2 second exposures at 4 second intervals. You get holes when you use the camera's built in "dark frame subtraction" feature on long exposures. Nikon calls this "long exposure noise reduction", and it engages on exposures over 1 second (most cameras) or 4 seconds (D200 and D80, of memory serves). When long exposure noise reduction is on, the camera "goes away" after each exposure for a length of time equal to the exposure length. So, there's a 2 second freeze after a 2 second exposure, and you end up with one 2 sec exposure every 4 seconds, and a 2 second "hole" between the exposures. Canon and Fuji cameras are more liberal about dark frame subtraction.
For example, if you're trying to make cotton candy waterfalls, you can tolerate holes. You can skip the neutral density filter, expose normally (say 1/100 sec at f16, ISO 100) and shoot sixty images, one every two seconds for two minutes, and get a wonderful waterfall. You could also use a neutral density filter, still shoot at f16, shoot 1 second exposures every 2 seconds for 2 minutes, and get pretty much the same results.
If you're trying to do a 10 minute shot of a building to "blur away" people, you need to shoot without holes, because 1/60 sec exposures at 2 second intervals leaves you a bunch of superimposed still people. If you're trying to make star trails, you should also shoot without holes. Sixty images, one every minute for an hour, with 2 second exposures, will give you dotted lines across the sky, instead of star trails.
Once you've made the decision about holes, compose your image, and lock the camera down on a very solid tripod. Use manual exposure mode, because you don't want the camera changing either aperture or exposure time during the course of the stack. If you're using an ND filter, compose without it, but measure exposure with it on.
If you're doing a daylight exposure, expose so that every frame in the stack will be properly exposed (maybe underexpose slightly to protect highlights). If you're shooting under dim light, you'll probably set an underexposure. This will also determine the number of shots in the stack. If you're doing a building at twilight and the camera says you're 3 stops under at 2 seconds, f8, then a shot of eight 2 second frames won't look all that different from a single 16 second exposure. 16 frames will show some enhanced tonality and shadow detail, and 32 frames will show greatly enhanced shadow detail. The rule of thumb is that twice the length of a "proper" exposure will give you noticeable enhancement, and there's about a full stop of added enhancement for every time you double the number of frames past that.
Then batch convert the raw files to TIFF, and feed them to the image stacking program. Don't just try to average them in PotatoeShop, because it can't align images, and most cameras cannot shoot images in accurate pixel alignment, even on a great tripod. For RegiStack, pick a "typical" image, find a few stationary objects (rocks, mountains, tree trunks, etc) and select them as alignment points. Then run the stacker, and save the result.
Viola!
First, you probably need image stacking software. I recommend AstroStack (about $100) or RegiStack (free). Both programs are capable of automatic, sub pixel alignment.
Then you need to decide if the image needs pretty much continuous sampling or if it can tolerate "holes". By holes, I mean whether each exposure has to be right on the tail of another, like two second exposures at two second intervals, or whether there can be gaps, like 1/60 sec exposures at 2 second intervals, or 2 second exposures at 4 second intervals. You get holes when you use the camera's built in "dark frame subtraction" feature on long exposures. Nikon calls this "long exposure noise reduction", and it engages on exposures over 1 second (most cameras) or 4 seconds (D200 and D80, of memory serves). When long exposure noise reduction is on, the camera "goes away" after each exposure for a length of time equal to the exposure length. So, there's a 2 second freeze after a 2 second exposure, and you end up with one 2 sec exposure every 4 seconds, and a 2 second "hole" between the exposures. Canon and Fuji cameras are more liberal about dark frame subtraction.
For example, if you're trying to make cotton candy waterfalls, you can tolerate holes. You can skip the neutral density filter, expose normally (say 1/100 sec at f16, ISO 100) and shoot sixty images, one every two seconds for two minutes, and get a wonderful waterfall. You could also use a neutral density filter, still shoot at f16, shoot 1 second exposures every 2 seconds for 2 minutes, and get pretty much the same results.
If you're trying to do a 10 minute shot of a building to "blur away" people, you need to shoot without holes, because 1/60 sec exposures at 2 second intervals leaves you a bunch of superimposed still people. If you're trying to make star trails, you should also shoot without holes. Sixty images, one every minute for an hour, with 2 second exposures, will give you dotted lines across the sky, instead of star trails.
Once you've made the decision about holes, compose your image, and lock the camera down on a very solid tripod. Use manual exposure mode, because you don't want the camera changing either aperture or exposure time during the course of the stack. If you're using an ND filter, compose without it, but measure exposure with it on.
If you're doing a daylight exposure, expose so that every frame in the stack will be properly exposed (maybe underexpose slightly to protect highlights). If you're shooting under dim light, you'll probably set an underexposure. This will also determine the number of shots in the stack. If you're doing a building at twilight and the camera says you're 3 stops under at 2 seconds, f8, then a shot of eight 2 second frames won't look all that different from a single 16 second exposure. 16 frames will show some enhanced tonality and shadow detail, and 32 frames will show greatly enhanced shadow detail. The rule of thumb is that twice the length of a "proper" exposure will give you noticeable enhancement, and there's about a full stop of added enhancement for every time you double the number of frames past that.
Then batch convert the raw files to TIFF, and feed them to the image stacking program. Don't just try to average them in PotatoeShop, because it can't align images, and most cameras cannot shoot images in accurate pixel alignment, even on a great tripod. For RegiStack, pick a "typical" image, find a few stationary objects (rocks, mountains, tree trunks, etc) and select them as alignment points. Then run the stacker, and save the result.
Viola!
